ELISA detection of IL-1 beta in human sera needs independent confirmation. False positives in hospitalized patients

The detection of cytokines in human sera has become an accepted index of disease activity in various diseases, including sepsis. However, little attention has been paid to the specificity of these measurements. Using a sensitive sandwich enzyme-linked immunoassay (ELISA), we studied IL-1 beta detectability in sera from 419 serum samples randomly obtained from our clinical laboratory. In initial studies, 6.7% of samples were positive (n = 24, 0.1 to 1.0 ng/ml, and n = 5, 1 to 80 ng/ml). However, attempts to further characterize positive samples revealed that over 90% were falsely positive. For example, samples fractionated on Sephadex G-75 demonstrated IL-1 beta "detectability" near the void volume, and negative samples, spiked with rIL-1 beta, eluted at approximately 17 kD. To determine if this detectability was due to heterophillike antibodies, 23 of 29 "positive" samples were retested in the presence of nonimmune mouse serum. Only 2 of 23 previously positive samples were still positive. Importantly, mouse serum had no effect upon normal human serum spiked with rIL-1 beta. Furthermore, blinded samples sent to a reference laboratory also demonstrated false positive IL-1 beta detection in selected samples. Taken together, these data demonstrate that the presence of nonspecific immunoactivity in sera may confound cytokine assays of human biologic material and suggest that, when possible, a second means of confirming ELISA-positive samples be used.